为实现钢箱梁外表面特殊复杂环境下涂层病害的智能检测、精准感知与识别，开展了机器人、多维原位无损检测技术与病害识别软件的技术集成研究，通过研制磁吸附爬壁机器人，开发高清相机模组、高光谱仪、 X射线衍射检测仪等便携式检测装备和搭建钢结构表面病害图像识别专家决策系统，成功研制了基于磁吸附爬壁机器人的钢箱梁表面涂层病害智能检测技术。结果表明：在港珠澳大桥实桥测试中，该技术关键设施表面巡检覆盖范围为 91. 25%，典型缺陷智能识别率为 90. 3%，定位精度为 7 mm，可识别线状缺陷宽度 ≥0. 104 mm，可识别斑状缺陷面积 ≥0. 93 mm×1. 02 mm，满足了大桥钢箱梁外表面高效、准确、安全检测的迫切需求。

In order to realize the intelligent detection，accurate perception and identification of coating defects on the outer surface of steel box girders in a special complexenvironment，we developed the robot technology，multi-dimensional in-situ non-destructive inspection and defects identification software integration. In this work，a magnetic adsorption wall-climbing robot technology was studied，the portable inspection equipment such as high-definition camera modules，hyper-spectrometers，and X-ray diffraction detector were developed，and an expert decision-making system for image recognition of steel structuresurface defects was built. It realized an intelligent detection technology for steel structures surface coating defects. The results showed that，in the actual bridge test of the Hong Kong-Zhuhai-Macao Bridge，this technology achieved a surface inspection coverage of 91. 25% of key facilities，an intelligent identification rate of typical defects of 90. 3%，a positioning accuracy of 7 mm，and the ability to identify linear defects with a width of ≥0. 104 mm. Thepatchy defect area was ≥0. 93 mm×1. 02 mm，which met the urgent need for efficient，accurate and safe inspection of the outer surface of the steel box girder for the bridge.